公开(公告)号：US20190340511A1

公开(公告)日：2019-11-07

申请号：US16447216

申请日：2019-06-20

Applicant: Intel Corporation

Inventor： Sofiane Yous ,  Hamza Yous

IPC: G06N3/08 ,  G06N5/04 ,  G06N3/04 ,  G06F17/15

Abstract: Systems, methods, computer program products, and apparatuses to transform a weight space of an inference model to increase the compute efficiency of a target inference platform. A density of a weight space can be determined, and a transformation parameter derived based on the determined density. The weight space can be re-ordered based on the transformation parameter to balance the compute load between the processing elements (PEs) of the target platform, and as such, reduce the idle time and/or stalls of the PEs.

公开(公告)号：US20240127031A1

公开(公告)日：2024-04-18

申请号：US18394307

申请日：2023-12-22

Applicant: Intel Corporation

Inventor： Hamza Yous ,  Ian Hunter ,  Alessandro Palla

IPC: G06N3/042 ,  G06N3/08

CPC classification number: G06N3/042 ,  G06N3/08

Abstract: A graph neural network (GNN) model is used in a scheduling process for compiling a deep neural network (DNN). The DNN, and parameter options for scheduling the DNN, are represented as a graph, and the GNN predicts a set of parameters that is expected to have a low cost. Using the GNN-based model, a compiler can produce a schedule for compiling the DNN in a relatively short and predictable amount of time, even for DNNs with many layers and/or many parameter options. For example, the GNN-based model reduces the overhead of exploring every parameter combination and does not exclude combinations from consideration like prior heuristic-based approaches.